osg'05 grass workshop slides - open source gis

Mar

kus

Net

eler

200

5

GRASS Workshop

Open Source Geospatial ’05 Conference

presented by

Markus Neteler – Kristen Perry

M. Netelerneteler at itc ithttp://mpa.itc.it

ITC-irst, Povo (Trento), Italy

K. Perrykperry at spatialfocus com

http://www.spatialfocus.com

USA

Mar

kus

Net

eler

200

5

GRASS: Geographic Resources Analysis Support System

Free Software GIS (“software libero”):

GRASS master Web site is in Italy: http://grass.itc.it

Portable: Versions for GNU/Linux, MS-Windows, Mac OSX, SUN, etc

Programming: Programmer's Manual on Web site (PDF, HTML), generated weekly. Code is documented in source code files (doxygen)

Sample data

Mailing lists in various languages

Commercial support

Mar

kus

Net

eler

200

5


Brief Introduction – Development and System Requirements

Developed since 1984, always Open Source, since 1999 under GNU General Public License

Written in C programming language, portable code (32/64bit) International development team, since 2001 coordinated at ITC-irst Distributed as source code, precompiled binaries for various

platforms, CDROM

GNU/Linux

MacOSX

MS-Windows

iPAQ


The new features of GRASS 6:

- New topological 2D/3D vector engine

- Support for vector analysis

- Attributes managed in an SQL-based DBMS

- New display manager and GRASS interface in QGIS

- Enhanced NVIZ 3D vector tool

- Integrated with GDAL/OGR libraries

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.


GRASS startup screen

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.

Creating a location from an EPSG code


Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.


The default GUI is the GIS / Display Manager (d.m)

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.


NVIZ: Perspective 2D Maps

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.


QGIS geodata browser: integrating GRASS functionality

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.


JAVAGRASS

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.


GRASS Command Overview

prefix function class type of command example

d.* display graphical output d.rast: views raster mapd.vect: views vector map

db.* database database management db.select: selects value(s) from table

g.* general general file operations g.rename: renames map

i.* imagery image processing i.smap: image classifier

ps.* postscript map creation in Postscript format ps.map: map creation

r.* raster raster data processing r.buffer: buffer around raster featuresr.mapcalc: map algebra

r3.* voxel raster voxel data processing r3.mapcalc: volume map algebra

v.* vector vector data processing v.overlay: vector map intersections

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.


Online help: Help button and g.manual

Kri

sten

Per

ry -

Spa

tial

Foc

us, I

nc.

Help can be found at different levels:

– Launching a GRASS command without parameter (in most cases) opens a graphical window:

<command> e.g., d.rastAt the bottom a HELP button is provided.

– Flags and parameters of a GRASS command you get with:<command> -help e.g., d.rast -help

– The manual page in a web browers you get with:g.manual <command> e.g., g.manual d.rast

– The manual page in MAN style you get with:g.manual -m <command> e.g., g.manual -m d.rast

GRASS: Geographic Resources Analysis Support SystemK

rist

en P

erry

- S

pati

al F

ocus

, Inc

.

Section 2: Introduction to GRASS


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

SD

Spearfish (SD) sample data location


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

/path/to/grass/locations/

here: /usr/local/osgis/grass_data


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

1.

2.

3.

Adding the elevation.dem raster map


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

1.

2.

3.

Adding the 'roads' vector map


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

What should display in the monitor


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

An approximate view of your saved region


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

NVIZ


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

QGIS


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

Creating a paper map


rist

en P

erry

- S

pati

al F

ocus

, Inc

.

To close GRASS

First close the QGIS window and display manager.

Type exit in the terminal to leave GRASS.

Monitors are closed automatically.

Mar

kus

Net

eler

200

5

Outline of part II a

3 Working with own data - Import/Export/Creating Locations

➢ Import of TIGER 2000 and LANDSAT-7 data

➢ Creating a new location external data files

➢ Creating from EPSG code/interactively a new location

http://mpa.itc.it/markus/mum3/

Mar

kus

Net

eler

200

5


Database: contains all GRASS data

Each GRASS project is organized in a „Location“ directory with subsequent„Mapset(s)“ subdirectories:

• Location: contains all spatial/attribute data of a geographically defined region (= project area)

• Mapset(s): used to subdivide data organization e.g. by user names, subregions or access rights (workgroups)

• PERMANENT: The PERMANENT mapset is a standard mapset which contains the definitions of a location. May also contain general cartography as it is visible to all users

Multi-User support: multiple users can work in a single location usingdifferent mapsets. Access rights can be managed per user. No user canmodify/delete data of other users.

Location and Mapset: “GRASS speech”

Mar

kus

Net

eler

200

5


Example for Location and Mapsets

/home/user/grassdata

/europa

/hannover

/world

histdblncoor

sidx

topo

MapsetLocation

/city

GRASS Database

/prov_trentino /trento

/silvia

Geometry and attribute data

streets

parks

lakes

poi

streets.dbf

parks.dbf

poi.dbf

lakes.dbf

fcell

hist

colr

cell_misc

cellhd

cell

cats

vector

dbf

Mar

kus

Net

eler

200

5

Import of TIGER 2000 vector maps

Please unpack the tiger2000_latlong_nad83.tar.gz file

To match the GRASS Spearfish sample dataset definitions (UTM zone 13N, NAD27/Clarke66; EPSG code 26713), we'llreproject them with OGR:

ogr2ogr -s_srs epsg:4269 -t_srs epsg:26713 \tgr46081lkA_UTM13_nad27.shp tgr46081lkA.shp

Reproject the maps tgr46081lkA... (roads) and tgr46081lkH... (hydro)

Now start GRASS again: grass60with Spearfish location

The import of the roads and the hydro maps is done with v.in.ogr(dsn: data source name is the reprojected input SHAPE file)

View the maps with d.vect or qgis

Mar

kus

Net

eler

200

5

Import of LANDSAT-7 Erdas/Img raster maps 1/2

0.4 0.6 0.8 1.21.0 1.4 1.6 1.8 2.0 2.2 2.4 2.6

-2e

xtr

ate

rr.

so

lar

radi

atio

n [

Wm

]

100

80

60

40

20

0

TM1 TM3TM4TM2 TM5 TM7

2000

1500

1000

500

0

Re

lativ

e s

pe

ctra

l se

nsi

tivity

[%

]

Wave length [micrometer]

A LANDSAT-7 scene has been prepared (reprojected, spatially subset):

- spearfish_landsat7_NAD27_vis_ir.img: TM10,TM20,TM30 (blue, green, red), TM40 (NIR), TM50, TM70 (MIR)- spearfish_landsat7_NAD27_tir.img: TM62 (TIR low gain), TM62 (TIR high gain)- spearfish_landsat7_NAD27_pan.img: TM80 (panchromatic)

Solar spectrum and LANDSAT channels (thermal channel 6 not shown)

Nete

ler/

Mit

aso

va 2

00

2

Mar

kus

Net

eler

200

5

Import of LANDSAT-7 Erdas/Img raster maps 2/2

The import is done with r.in.gdal:

r.in.gdal -e in=spearfish_landsat7_NAD27_vis_ir.img out=tm # To keep the numbering right, we rename tm.6 to the # correct number tm.7:g.rename rast=tm.6,tm.7

r.in.gdal -e in=spearfish_landsat7_NAD27_tir.img out=tm6

r.in.gdal -e in=spearfish_landsat7_NAD27_pan.img out=pan

Generate a RGB composite on the fly (zoom to map first):

g.region rast=tm.1 -pd.rgb b=tm.1 g=tm.2 r=tm.3

You should see the Spearfish area in near-natural colors.

Mar

kus

Net

eler

200

5

Creating new GRASS locations

Both r.in.gdal and v.in.ogr offer a location= parameter tocreate a new location from the import dataset's metadata

Example:r.in.gdal -e in=spearfish_landsat7_NAD27_tir.img out=tm6 location=utm13

Launching GRASS (again) permits to

create a new location from EPSG code

create a new location interactively

See the workshop handout for details

Mar

kus

Net

eler

200

5

Outline of part II b

4 Raster map analysis

➢ DEM analysis

➢ Raster map algebra

➢ Geocoding of scanned map

➢ Volume data processing


Mar

kus

Net

eler

200

5

Raster data analysis: Slope and aspect from DEM

Calculating slope and aspect from a DEM

# First we reset the current GRASS region settings to the input map: g.region rast=elevation.dem -p

r.slope.aspect el=elevation.dem as=aspect_30m sl=slope_30m

d.rast aspect_30md.rast.leg slope_30m

Note: horizontal angles are counted counterclockwise from the East

Slopes are calculated by default in degrees

Also curvatures can be calculated

0° East360°

+

Mar

kus

Net

eler

200

5

Raster data analysis: Geomorphology

DEM: r.param.scale

DEM: 30mSize: 7x7

# zoom to the input map:g.region rast=elevation.dem -p

# generalize with size parameterr.param.scale elevation.dem out=morph \ param=feature size=21

# with legendd.rast.leg morph

# view with aspect/shade map (or QGIS)d.his h=morph i=aspect.10m

Mar

kus

Net

eler

200

5

Raster data analysis: Water flows - Contributing area

Topographic Index: ln(a/tan(beta)) g.region rast=elevation.10m -p# zoom to smaller area d.zoom

r.topidx in=elevation.10m out=ln_a_tanB d.rast ln_a_tanB d.vect streams col=yellow nviz elevation.10m col=ln_a_tanB

Mar

kus

Net

eler

200

5

Raster data analysis: further methods

Additional DEM analysis modules:- depression areas can be filled with r.fill.dir- flowlines can be calculated with r.flow- trace a flow through a DEM: r.drain- watershed analysis can be done with r.watershed and r.terraflow- cost surfaces: r.cost

Energy:- cast shadows, astronomical calculations of sun position: r.sunmask- energy budget: r.sun

Line of sight:- viewsheds can be generated with: r.los

Interpolation methods- 2D inverse distance weighted: v.surf.idw- 2D from contour lines: r.surf.contour- 2D bilinear: r.bilinear- 2D regularized splines with tension (with cross validation): v.surf.rst- 3D regularized splines with tension (with cross validation): v.vol.rst- 2D/3D kernel densities: v.kernel

- via R-stats: kriging, predictive models etc

Mar

kus

Net

eler

200

5

Raster map algebra

A powerful raster map algebra calculator is r.mapcalcSee for functionality:

g.manual r.mapcalc &

With a simple formula we filter all pixels with elevation higher than 1500m from the Spearfish DEM:

r.mapcalc "elev_1500 = if(elevation.dem > 1500.0, elevation.dem, null())”d.rast elev_1500

d.rast aspectd.rast -o elev_1500

Mar

kus

Net

eler

200

5

Volume map processing: Demo

GRASS was enhanced to process and visualize Volumes(consisting of 3D voxels)

Functionality:

3D import/export

3D Regularized Splines with Tension interpolation

3D map algebra

NVIZ volume visualization: Isosurfaces and Profiles

Mar

kus

Net

eler

200

5

Outline of part II d

5 Image processing

➢ Image classification

➢ Image fusion with Brovey transform

➢ Calculating a degree Celsius map from the LANDSAT thermal channel


Mar

kus

Net

eler

200

5

Import of LANDSAT-7 Erdas/Img

Unsupervised & Supervised Image Classification

➢ classification methods in GRASS:

➢ all image data must be first listed in a group (i.group)

➢ See handout for unsupervised classification example

radiometric, radiometric, supervised radio- and geometricunsupervised supervised

Preprocessing i.cluster i.class (monitor) i.gensig (maps) i.gensigset (maps) Computation i.maxlik i.maxlik i.maxlik i.smap

Image Classification

Mar

kus

Net

eler

200

5


Image classification

© 2

00

4 G

DF

Hannover,

Germ

any

Biotope monitoring from digital aerial cameras (HRSC-X and DMC)

SMAP Classifier of GRASS

GRASS supports Image geocoding and

ortho-rectification Analysis of aerial and satellite data Time series analysis

Mar

kus

Net

eler

200

5

Image fusion: Brovey transform

The earlier imported LANDSAT-7 scene wil be used to perform image fusionof the channels 2 (red), 4 (NIR), and 5 (MIR):

g.region -dpi.fusion.brovey -l ms1=tm.2 ms2=tm.4 ms3=tm.5 pan=pan out=brovey

# zoom to fused channelg.region -p rast=brovey.red

# color composite:r.composite r=brovey.red g=brovey.green b=brovey.blue n out=tm.brovey

d.rast tm.brovey

nviz elevation.10m col=tm.brovey

# Increase visual resolution in NVIZ # with Panel -> Surface # -> Polygon resolution# (lower! the value)

Mar

kus

Net

eler

200

5

TM61: Conversion of temperature first to Kelvin, then to degree Celsius

g.region rast=tm6.1 -p

#DN: digital numbers (coded temperatures) r.info -r tm6.1 min=131 max=175

# Conversion of DN to spectral radiances: r.mapcalc "tm61rad=((17.04 - 0.)/(255. - 1.))*(tm6.1 - 1.) + 0." r.info -r tm61rad min=8.721260 max=11.673071

# Conversion of spectral radiances to absolute temperatures (Kelvin):# T = K2/ln(K1/L_l + 1)) r.mapcalc "temp_kelvin=1260.56/(log (607.76/tm61rad + 1.0))" r.info -r temp_kelvin min=296.026722 max=317.399879

Recalibrating the LANDSAT-7 thermal channel 1/2

Mar

kus

Net

eler

200

5

TM61: ... conversion to degree Celsius

# We currently have the land surface temperature map in Kelvin.# Conversion to degree Celsius: r.mapcalc "temp_celsius=temp_kelvin - 273.15" r.info -r temp_celsius min=22.876722 max=44.249879

# New color table:r.colors temp_celsius col=rules << EOF-10 blue15 green25 yellow35 red50 brownEOF

d.rast.leg temp_celsius

g.region rast=elevation.dem -pnviz elevation.dem col=temp_celsius

Note: Land surface temperatures are not air temperatures!

LANDSAT passes at around 9:30 local time

Recalibrating the LANDSAT-7 thermal channel 2/2

Mar

kus

Net

eler

200

5

Outline of part II e

6 Working with vector data

➢ Vector map import

➢ Attribute management

➢ Buffering

➢ Extractions, selections, clipping, unions, intersections

➢ Conversion raster-vector and vice verse

➢ Digitizing in GRASS and QGIS

➢ Working with vector geometry


Mar

kus

Net

eler

200

5

GRASS 6 Vector data

Vector geometry types

Point Centroid Line Boundary Area (boundary + centroid) face (3D area) [kernel (3D centroid)] [volumes (faces + kernel)]

Geometry is true 3D: x, y, z

Node

Node

Vertex

Vertex

Segme

ntSeg

ment

Segment

Node

Boundary

Vertex

Vertex

Vertex

Vertex

Centroid

Area

Line

Faces

Mar

kus

Net

eler

200

5

Raster-Vector conversion – extraction 1/2

Extraction of residential areas from raster landuse map

# look at the landuse/landcover map with legend: g.region rast=landcover.30m -p d.erase d.rast.leg -n landcover.30m

# Automated vectorization of the landuse/landcover map: r.to.vect -s landcover.30m out=landcover30m feature=area

# see attribute table: v.db.connect -p landcover30m v.db.select landcover30m

Mar

kus

Net

eler

200

5

Raster-Vector conversion – extraction 2/2

Extraction of residential areas from raster landuse map

# generate list of unique landuse/landcover types: v.db.select landcover30m | sort -t '|' -k2 -un

#display selected categories: d.erase d.vect landcover30m where="value=21 or value=22" fcol=orange

# Extract residential area into a new vector map: v.extract landcover30m out=residential where="value=21 or value=22"

d.frame -ed.vect residential fcol=orange \ type=aread.vect roadsd.barscale -mt

Mar

kus

Net

eler

200

5

Creating/modifying vector maps

Digitizing in GRASS

21

g.region rast=landcover.30m -pv.digit -n map=cities \ bg="d.rast landcover.30m"

1. define table set snapping threshold2. start digitizing

Alternative:QGIS digitizer

Mar

kus

Net

eler

200

5

Vector map clipping

TIGER data: roads in urban areas

# import urban areas (from package tiger2000_UTM13_nad27.tar.gz): v.in.ogr dsn=UA_46081_UTM13_nad27.shp out=urban_areas d.vect urban_areas type=area

# display roads: d.vect roads

# extract all roads within the urban areas: v.select ain=roads bin=urban_areas out=urban_roads d.vect urban_roads col=red

Mar

kus

Net

eler

200

5


•In GRASS an area polygon is defined by a boundary + a centroid.•Lines can be a (poly)line or a boundary.•

•Vector types can be changed by v.type/v.build.polyline such as•point centroid•3D point kernel (3D centroid)•line polyline•line boundary•3D area face

Boundaries + centroids Lines + centroids

Changing vector types

Mar

kus

Net

eler

200

5

Outline of part II f

7 Vector networking

➢ Overview

➢ Shortest path analysis


Mar

kus

Net

eler

200

5

Vector network analysis methods

Available methods:

find shortest path along vector network - road navigation

find optimal round trip visiting selected nodes (Traveling salesman problem) - delivering of goods

find optimal connection between nodes (Minimum Steiner tree) - ADSL network

subdivide a network in subnetworks (iso distances) - how far can I go from a node in all directions

find subnetworks for set of nodes (subnet allocation) - “catchment area” for fire brigade etc

Mar

kus

Net

eler

200

5

Vector network analysis methods

Vector network with one way roads

Generic vector directions One attribute column for each direction Value -1 closes direction (for one way streets)

drawn inps.map

Street directionopenclosed

Mar

kus

Net

eler

200

5

Vector networking

Shortest path with d.path

d.vect roadsd.path roads

# or:# v.net.path

Mar

kus

Net

eler

200

5

Outline of part II g

8 GRASS and geostatistics

➢ R-stats/GRASS interface

Closure of the Workshop


Mar

kus

Net

eler

200

5

R-stats is a powerful statistical language

Spatial extentions available for all kinds of geostatistics, spatial pattern analysis, time series etc

Interface to exchange raster and point data between GRASS and R-stats

Rdbi: connects R-stats to PostgreSQL

GRASS

PostgreSQL

Spatial data

Tables

GeostatisticsPredictive Models

Portable Applications: - GNU/Linux - MS-Windows - MacOSX- ...

http://www.r-project.orghttp://grass.itc.it/statsgrass/

GRASS/R-stats interface - R-stats/PostgreSQL interface

Mar

kus

Net

eler

200

5

R statistical language

Web site and CRAN: http://www.r-project.org http://cran.r-project.org

Object oriented statistical language, a “S” dialect. Examples: R > 1 [1]

> 1+1 [1] 2

# assignment into object: > x <- 1+1 > x [1] 2

> q() Save workspace image [y/n/c] n

GRASS/R-stats interface

Mar

kus

Net

eler

200

5

GRASS 6 and R statistical language

grass60

#reset region:g.region rast=elevation.dem -p

#in GRASS start R:Rlibrary(spgrass6)

#load GRASS environment into R:G <- gmeta6()

#see environment metadata:str(G)

# Now R is ready for GRASS data analysis.


Mar

kus

Net

eler

200

5

GRASS 6 and R statistical language (cont'd)

# get online help:?readCELL6sp

# get full help:help.start()

# load GRASS DEM into R:elev <- readCELL6sp("elevation.dem")

# show map metadata:summary(elev)

# show map:image(elev, col=terrain.colors(20))

# leave R:q()

# you have the option to save the current R state to local# file when leaving the program.


Mar

kus

Net

eler

200

5

Closure of the Workshop

Thanks for your interest and your attention!


M. Netelerneteler at itc ithttp://mpa.itc.it

http://www.gdf-hannover.de

ITC-irst, Povo (Trento), ItalyGDF Hannover, Germany

K. Perrykperry at spatialfocus com

http://www.spatialfocus.com

USA

osg'05 grass workshop slides - open source gis

Documents